Head-to-head comparison

GPT-5.4 mini vs. Gemini 3.8 Flash

Compare published benchmark results from matching versions and cohorts, API pricing, and technical specifications. Any documented test setup differences remain visible.

shared result series: 212
distinct benchmarks: 22
latest retrieval: 2026-10-15

First modelSecond model

Specifications

General specifications

Release date, availability, context window, architecture, and published parameter counts at a glance. Undisclosed details remain marked as unavailable.

Specifications for GPT-5.4 mini and Gemini 3.8 Flash
Attribute	GPT-5.4 mini	Gemini 3.8 Flash
Provider	OpenAI	Google
Released	Mar 17, 2026	Sep 2, 2026
Availability	Active	Active
Model type	Proprietary	Proprietary
Parameters	Not published	Not published
Architecture	Not published	Not published
Context window	400,000 tokens	1,048,576 tokens
Knowledge cutoff	August 31, 2025	March 2026
Technical sources	Model	Model, Knowledge cutoff

Pricing

API pricing compared

The table compares official standard API rates per 1M tokens, including published cache pricing. Batch, fast, priority, regional surcharges, tool calls, and cloud platform pricing are excluded.

API pricing for GPT-5.4 mini and Gemini 3.8 Flash
Price type	GPT-5.4 mini	Gemini 3.8 Flash
API input	$0.75	$0.75
API output	$4.5	$3.75
Cache read	$0.075	$0.075
Cache write (5 min.)	Not listed	Not listed
Cache write (1 hr.)	Not listed	Not listed
Price verified	09/22/2026 Source	09/22/2026 Source

GPT-5.4 miniGemini 3.8 Flash

API inputUSD per 1M tokens, base tier

GPT-5.4 mini$0.75

Gemini 3.8 Flash$0.75

API outputUSD per 1M tokens, base tier

GPT-5.4 mini$4.5

Gemini 3.8 Flash$3.75

Cache readUSD per 1M tokens, base tier

GPT-5.4 mini$0.075

Gemini 3.8 Flash$0.075

Key to the marks:better valuebetter, but not conclusivetie

Pricing by provider

Direct developer pricing and provider endpoint tariffs routed through OpenRouter are listed separately. Prices are in USD per 1M tokens. Context-dependent direct tiers plus regional, Flex, Priority, and other OpenRouter tariffs remain individually identifiable.

Provider pricing for GPT-5.4 mini and Gemini 3.8 Flash
Model	Provider	Purchase route and tariff	Input per 1M tokens	Output per 1M tokens	Source
GPT-5.4 mini	OpenAI	Direct from developerStandard API	$0.75	$4.5	Source09/22/2026
	Azure	Via OpenRouterazure	$0.75	$4.5	Source10/17/2026
	Azure	Via OpenRouterazure/us	$0.825	$4.95	Source10/17/2026
	OpenAI	Via OpenRouteropenai	$0.75	$4.5	Source10/17/2026
	OpenAI	Via OpenRouteropenai/fast	$1.5	$9	Source10/17/2026
	OpenAI	Via OpenRouteropenai/flex	Lowest listed price: $0.375	Lowest listed price: $2.25	Source10/17/2026
Gemini 3.8 Flash	Google	Direct from developerStandard API	$0.75	$3.75	Source09/22/2026
	Google	Via OpenRoutergoogle-vertex/us	$0.825	$4.125	Source10/17/2026
	Google AI Studio	Via OpenRoutergoogle-ai-studio	$0.75	$3.75	Source10/17/2026
	Google AI Studio	Via OpenRoutergoogle-ai-studio/flex	Lowest listed price: $0.375	Lowest listed price: $1.875	Source10/17/2026
	Google AI Studio	Via OpenRoutergoogle-ai-studio/priority	$1.35	$6.75	Source10/17/2026

Performance

Shared benchmarks

Only published results with the same benchmark version, task, metric, and comparison cohort are paired. Different reasoning levels, agents, harnesses, or output limits appear directly in the table.

Capability profile from matched benchmarks

Each axis averages directly matched benchmark families. A value of 100 means the stronger value within this pair, not a universal quality score.

GPT-5.4 mini

Gemini 3.8 Flash

Coding: 92.9 / 100 (2 benchmark families)

Reasoning: 97 / 100 (1 benchmark family)

Mathematics: 95.4 / 100 (1 benchmark family)

Knowledge: 95.8 / 100 (1 benchmark family)

Finance: 98.6 / 100 (1 benchmark family)

Legal: 97.6 / 100 (1 benchmark family)

Multimodal: 96.2 / 100 (1 benchmark family)

Health: 96.1 / 100 (1 benchmark family)

40 matched rows from 3 benchmark families were considered. The radar shows 8 of 9 comparable categories.

GPT-5.4 miniGemini 3.8 Flash

Code MigrationOverall. Higher is better. Test setup not documented as identical

GPT-5.4 mini12.936 %

Gemini 3.8 Flash36.546 %

EMBOverall. Higher is better. Test setup not documented as identical

GPT-5.4 mini45.426 %

Gemini 3.8 Flash72.196 %

Finance Agent (v2)Overall. Higher is better. Test setup not documented as identical

GPT-5.4 mini45.36 %

Gemini 3.8 Flash61.435 %

GPQA DiamondOverall. Higher is better. Test setup not documented as identical

GPT-5.4 mini83.08 %

Gemini 3.8 Flash94.444 %

Harvey's Legal Agent BenchmarkOverall · Task fully resolved. Higher is better. Test setup not documented as identical

GPT-5.4 mini0 %

Gemini 3.8 Flash10 %

Legal Research BenchOverall · All-pass. Higher is better. Test setup not documented as identical

GPT-5.4 mini12.5 %

Gemini 3.8 Flash38.942 %

LiveCodeBench v6Overall. Higher is better. Test setup not documented as identical

GPT-5.4 mini81.465 %

Gemini 3.8 Flash89.484 %

LMArena Code Arena WebDevOverall. Higher is better. No documented setup difference

GPT-5.4 mini1,397.568

Gemini 3.8 Flash1,584.77

LMArena Text, Style ControlOverall. Higher is better. No documented setup difference

GPT-5.4 mini1,447.149

Gemini 3.8 Flash1,496.319

LMArena Vision, Style ControlOverall. Higher is better. No documented setup difference

GPT-5.4 mini1,250.938

Gemini 3.8 Flash1,289.129

MMLU-ProOverall. Higher is better. Test setup not documented as identical

GPT-5.4 mini84.554 %

Gemini 3.8 Flash90.221 %

MMMU-ProOverall. Higher is better. Test setup not documented as identical

GPT-5.4 mini79.249 %

Gemini 3.8 Flash89.075 %

Key to the marks:better valuebetter, but not conclusivetie

Benchmark scores for GPT-5.4 mini and Gemini 3.8 Flash
Benchmark and task	GPT-5.4 mini	Gemini 3.8 Flash	Source
Code MigrationOverall Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	12.936 %	Higher measured value36.546 %	Vals AI Source details Data as of: 10/12/2026
EMBOverall Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	45.426 %	Higher measured value72.196 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)Overall Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	45.36 %	Higher measured value61.435 %	Vals AI Source details Data as of: 10/12/2026
GPQA DiamondOverall Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	83.08 %	Higher measured value94.444 %	Vals AI Source details Data as of: 09/01/2026
Harvey's Legal Agent BenchmarkOverall · Task fully resolved Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: task resolution rateScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	0 %	Higher measured value10 %	Vals AI Source details Data as of: 10/12/2026
Legal Research BenchOverall · All-pass Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: all-pass rateScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	12.5 %	Higher measured value38.942 %	Vals AI Source details Data as of: 10/12/2026
LiveCodeBench v6Overall Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	81.465 %	Higher measured value89.484 %	Vals AI Source details Data as of: 09/01/2026
LMArena Code Arena WebDevOverall Test details No documented difference in the test setupVersion: webdevMetric: WebDev Arena Bradley-Terry ratingScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-high, Reasoning: high, Harness: Arena Code Arena WebDev Gemini 3.8 Flash: Snapshot: gemini-3.8-flash-high, Reasoning: high, Harness: Arena Code Arena WebDevGPT-5.4 mini: Arena rating from 11,978 published pairwise votes.Gemini 3.8 Flash: Arena rating from 10,077 published pairwise votes.	1,397.568 95% confidence interval: ±6.522The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	Winner1,584.77 95% confidence interval: ±7.59The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
LMArena Text, Style ControlOverall Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-high, Reasoning: high, Harness: Arena Text, style control Gemini 3.8 Flash: Snapshot: gemini-3.8-flash-high, Reasoning: high, Harness: Arena Text, style controlGPT-5.4 mini: Arena rating from 63,461 published pairwise votes.Gemini 3.8 Flash: Arena rating from 35,670 published pairwise votes.	1,447.149 95% confidence interval: ±3.749The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	Winner1,496.319 95% confidence interval: ±4.654The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
LMArena Vision, Style ControlOverall Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-high, Reasoning: high, Harness: Arena Vision, style control Gemini 3.8 Flash: Snapshot: gemini-3.8-flash-high, Reasoning: high, Harness: Arena Vision, style controlGPT-5.4 mini: Arena rating from 25,733 published pairwise votes.Gemini 3.8 Flash: Arena rating from 3,792 published pairwise votes.	1,250.938 95% confidence interval: ±6.248The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	Winner1,289.129 95% confidence interval: ±10.62The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/09/2026
MMLU-ProOverall Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	84.554 %	Higher measured value90.221 %	Vals AI Source details Data as of: 09/01/2026
MMMU-ProOverall Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	79.249 %	Higher measured value89.075 %	Vals AI Source details Data as of: 09/01/2026
MortgageTaxOverall Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	63.514 %	Higher measured value65.342 %	Vals AI Source details Data as of: 09/01/2026
ProgramBenchOverall Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: fully resolved rateScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	0 %	Higher measured value1 %	Vals AI Source details Data as of: 10/12/2026
SAGEOverall Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 30,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 30,000 tokens	Higher measured value50.813 %	35.064 %	Vals AI Source details Data as of: 10/08/2026
SWE-bench VerifiedOverall Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	73 %	Higher measured value80 %	Vals AI Source details Data as of: 09/01/2026
Tax Agent BenchOverall Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	8.808 %	Higher measured value32.124 %	Vals AI Source details Data as of: 10/12/2026
TaxEval v2Overall Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	71.218 %	Higher measured value74.448 %	Vals AI Source details Data as of: 09/01/2026
Terminal-Bench 2.1Overall Test details Test setup not documented as identicalNo direct winnerVersion: 2.1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: high, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	54.682 %	Higher measured value81.273 %	Vals AI Source details Data as of: 09/28/2026
Terminal-Bench 4.0Overall Test details Test setup not documented as identicalNo direct winnerVersion: 4.0Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	2.525 %	Higher measured value19.697 %	Vals AI Source details Data as of: 10/13/2026
Vals IndexOverall Test details Test setup not documented as identicalNo direct winnerVersion: 2.1Metric: weighted index scoreScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	33.169 %	Higher measured value54.887 %	Vals AI Source details Data as of: 10/13/2026
Vibe Code Bench v1.1Overall Test details Test setup not documented as identicalNo direct winnerVersion: 1.1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Harness: OpenHands, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Harness: OpenHands, Temperature: 1, Output limit: 65,536 tokens	47.969 %	Higher measured value78.651 %	Vals AI Source details Data as of: 10/12/2026
Code MigrationCLI Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	8.2 %	Higher measured value25.871 %	Vals AI Source details Data as of: 10/12/2026
EMBComps Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	40.488 %	Higher measured value76.435 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)Adjustments Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	45.366 %	Higher measured value54.828 %	Vals AI Source details Data as of: 10/12/2026
GPQA DiamondFew-Shot CoT Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	84.848 %	Higher measured value94.949 %	Vals AI Source details Data as of: 09/01/2026
Harvey's Legal Agent BenchmarkAntitrust Competition · Criteria passed Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: criteria pass rateScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	52.536 %	Higher measured value87.681 %	Vals AI Source details Data as of: 10/12/2026
Legal Research BenchAdministrative / Regulatory · All-pass Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: all-pass rateScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	15.152 %	Higher measured value46.97 %	Vals AI Source details Data as of: 10/12/2026
LiveCodeBench v6Easy Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	97.826 %	Higher measured value99.068 %	Vals AI Source details Data as of: 09/01/2026
LMArena Code Arena WebDevWebdev Test details No documented difference in the test setupVersion: webdevMetric: WebDev Arena Bradley-Terry ratingScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-high, Reasoning: high, Harness: Arena Code Arena WebDev Gemini 3.8 Flash: Snapshot: gemini-3.8-flash-high, Reasoning: high, Harness: Arena Code Arena WebDevGPT-5.4 mini: Arena rating from 11,978 published pairwise votes.Gemini 3.8 Flash: Arena rating from 10,077 published pairwise votes.	1,397.568 95% confidence interval: ±6.522The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	Winner1,584.77 95% confidence interval: ±7.59The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
LMArena Text, Style ControlBusiness, management and finance Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-high, Reasoning: high, Harness: Arena Text, style control Gemini 3.8 Flash: Snapshot: gemini-3.8-flash-high, Reasoning: high, Harness: Arena Text, style controlGPT-5.4 mini: Arena rating from 12,861 published pairwise votes.Gemini 3.8 Flash: Arena rating from 6,520 published pairwise votes.	1,459.776 95% confidence interval: ±6.483The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	Winner1,479.887 95% confidence interval: ±8.136The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
LMArena Vision, Style ControlChinese Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-high, Reasoning: high, Harness: Arena Vision, style control Gemini 3.8 Flash: Snapshot: gemini-3.8-flash-high, Reasoning: high, Harness: Arena Vision, style controlGPT-5.4 mini: Arena rating from 1,662 published pairwise votes.Gemini 3.8 Flash: Arena rating from 182 published pairwise votes.	1,285.423 95% confidence interval: ±17.385The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	Winner1,379.729 95% confidence interval: ±43.521The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/09/2026
MMLU-ProBiology Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	91.353 %	Higher measured value94.979 %	Vals AI Source details Data as of: 09/01/2026
MortgageTaxNumerical Extraction Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	Higher measured value79.412 %	70.588 %	Vals AI Source details Data as of: 09/01/2026
ProgramBenchAlmost Resolved Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: almost-resolved rateScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	3 %	Higher measured value14 %	Vals AI Source details Data as of: 10/12/2026
SAGEAutomata Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 30,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 30,000 tokens	Higher measured value41.27 %	33.334 %	Vals AI Source details Data as of: 10/08/2026
SWE-bench Verified<15 min fix Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	85.567 %	Higher measured value89.691 %	Vals AI Source details Data as of: 09/01/2026
Tax Agent BenchControversy & Precedence Analysis Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	3.333 %	Higher measured value6.667 %	Vals AI Source details Data as of: 10/12/2026
TaxEval v2Correctness Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	Higher measured value63.042 %	62.469 %	Vals AI Source details Data as of: 09/01/2026
Terminal-Bench 2.1Easy Test details Test setup not documented as identicalNo direct winnerVersion: 2.1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: high, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	75 %	Higher measured value100 %	Vals AI Source details Data as of: 09/28/2026
Terminal-Bench 4.0Hardware Test details Test setup not documented as identicalNo direct winnerVersion: 4.0Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	0 %	Higher measured value26.667 %	Vals AI Source details Data as of: 10/13/2026
Vals IndexCode Migration Test details Test setup not documented as identicalNo direct winnerVersion: 2.1Metric: benchmark scoreScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	13.015 %	Higher measured value31.558 %	Vals AI Source details Data as of: 10/13/2026
Code MigrationCOBOL Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	27.143 %	Higher measured value68.571 %	Vals AI Source details Data as of: 10/12/2026
EMBDataroom Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	48.286 %	Higher measured value86.218 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)All-Pass Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	32.369 %	Higher measured value49.694 %	Vals AI Source details Data as of: 10/12/2026
GPQA DiamondZero-Shot CoT Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	81.313 %	Higher measured value93.939 %	Vals AI Source details Data as of: 09/01/2026
Harvey's Legal Agent BenchmarkAntitrust Competition · Task fully resolved Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: task resolution rateScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	No direct winner0 %	No direct winner0 %	Vals AI Source details Data as of: 10/12/2026
Legal Research BenchAdministrative / Regulatory · Weighted partial credit Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: weighted partial-credit scoreScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	69.72 %	Higher measured value87.613 %	Vals AI Source details Data as of: 10/12/2026
LiveCodeBench v6Hard Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	61.714 %	Higher measured value78 %	Vals AI Source details Data as of: 09/01/2026
LMArena Code Arena WebDevWebdev-html Test details No documented difference in the test setupVersion: webdevMetric: WebDev Arena Bradley-Terry ratingScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-high, Reasoning: high, Harness: Arena Code Arena WebDev Gemini 3.8 Flash: Snapshot: gemini-3.8-flash-high, Reasoning: high, Harness: Arena Code Arena WebDevGPT-5.4 mini: Arena rating from 1,436 published pairwise votes.Gemini 3.8 Flash: Arena rating from 1,451 published pairwise votes.	1,419.865 95% confidence interval: ±16.676The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	Winner1,620.104 95% confidence interval: ±17.559The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
LMArena Text, Style ControlChinese Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-high, Reasoning: high, Harness: Arena Text, style control Gemini 3.8 Flash: Snapshot: gemini-3.8-flash-high, Reasoning: high, Harness: Arena Text, style controlGPT-5.4 mini: Arena rating from 3,925 published pairwise votes.Gemini 3.8 Flash: Arena rating from 2,903 published pairwise votes.	1,473.388 95% confidence interval: ±10.187The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	Winner1,532.833 95% confidence interval: ±11.757The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/13/2026
LMArena Vision, Style ControlCreative writing vision Test details No documented difference in the test setupNo clear leadVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-high, Reasoning: high, Harness: Arena Vision, style control Gemini 3.8 Flash: Snapshot: gemini-3.8-flash-high, Reasoning: high, Harness: Arena Vision, style controlGPT-5.4 mini: Arena rating from 1,627 published pairwise votes.Gemini 3.8 Flash: Arena rating from 363 published pairwise votes.	1,226.475 95% confidence interval: ±15.662The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	No clear lead1,263.378 95% confidence interval: ±32.508The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.	LMArena Source details Data as of: 10/09/2026
MMLU-ProBusiness Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	89.734 %	Higher measured value93.79 %	Vals AI Source details Data as of: 09/01/2026
MortgageTaxSemantic Extraction Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	47.615 %	Higher measured value60.095 %	Vals AI Source details Data as of: 09/01/2026
ProgramBenchFully Resolved Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: fully resolved rateScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	0 %	Higher measured value1 %	Vals AI Source details Data as of: 10/12/2026
SAGECalculus Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 30,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 30,000 tokens	59.259 %	Higher measured value79.63 %	Vals AI Source details Data as of: 10/08/2026
SWE-bench Verified>4 hours Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	0 %	Higher measured value33.333 %	Vals AI Source details Data as of: 09/01/2026
Tax Agent BenchControversy & Precedence Analysis Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	32.174 %	Higher measured value59.001 %	Vals AI Source details Data as of: 10/12/2026
TaxEval v2Stepwise Reasoning Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	79.395 %	Higher measured value86.427 %	Vals AI Source details Data as of: 09/01/2026
Terminal-Bench 2.1Hard Test details Test setup not documented as identicalNo direct winnerVersion: 2.1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: high, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	34.444 %	Higher measured value72.222 %	Vals AI Source details Data as of: 09/28/2026
Terminal-Bench 4.0Media Test details Test setup not documented as identicalNo direct winnerVersion: 4.0Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	0 %	Higher measured value16.667 %	Vals AI Source details Data as of: 10/13/2026
Vals IndexEMB Test details Test setup not documented as identicalNo direct winnerVersion: 2.1Metric: benchmark scoreScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	45.426 %	Higher measured value72.196 %	Vals AI Source details Data as of: 10/13/2026
Code MigrationCode Quality Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: mean reviewer code-quality ratingScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	3.053 / 5	Higher measured value3.742 / 5	Vals AI Source details Data as of: 10/12/2026
EMBDCF Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.4 mini: Snapshot: gpt-5.4-mini-2026-03-17, Reasoning: xhigh, Output limit: 128,000 tokens Gemini 3.8 Flash: Reasoning: high, Temperature: 1, Output limit: 65,536 tokens	40.605 %	Higher measured value63.655 %	Vals AI Source details Data as of: 10/12/2026
Key to the marks:better valuebetter, but not conclusivetie

Our model tests

Our tests

SVG short test

One SVG prompt and fixed output settings. Captures are shown only with 0 reported reasoning tokens or documented lack of reasoning controls. Identical test conditions do not make the models' internal compute identical.

Shared test round: 4ae5d8eeb9a9

GPT-5.4 mini

Unavailable in this round

Gemini 3.8 Flash

Unavailable in this round

Earlier tests

Pelican on a bicycle

A community classic for free-form SVG drawing.

GPT-5.4 mini

Reasoning (requested): high

Current attempt

Run date: Oct 2, 2026
API provider: openrouter
Requested API model: openai/gpt-5.4-mini
API model reported in response: openai/gpt-5.4-mini
Reasoning setting: high
Pinned endpoint: OpenAI | openai/gpt-5.4-mini-20260317
Temperature: Provider default
Token limit including reasoning: 8,192 tokens
Test protocol: community-visual-tests-v1
Technical checks: Passed, not a quality rating
Response time: 13.5 s

Visually reviewed. This individual test does not establish overall model quality.

Earlier attempts (2)

Oct 1, 2026API request failedReasoning setting: high
Sep 7, 2026API request failedReasoning setting: high

Gemini 3.8 Flash

The response reached the token limit. No complete recording is available.

Reasoning (requested): high

Current attempt

Run date: Oct 5, 2026
API provider: openrouter
Requested API model: google/gemini-3.8-flash
API model reported in response: google/gemini-3.8-flash
Reasoning setting: high
Pinned endpoint: Google AI Studio | google/gemini-3.8-flash-20260902
Temperature: Provider default
Token limit including reasoning: 8,192 tokens
Test protocol: community-visual-tests-v1
Technical checks: Failed
Response time: 40.2 s

Not yet visually reviewed. A successful recording does not confirm correct physics or full compliance with the prompt.

Earlier attempts (3)

Oct 2, 2026API request failedReasoning setting: high
Oct 1, 2026API request failedReasoning setting: high
Sep 7, 2026API request failedReasoning setting: high

Prompt and test conditions

Generate an SVG of a pelican riding a bicycle

Token limit including reasoning: 8,192 tokens

Task origin (Simon Willison)

How to read this comparison

Winning one benchmark is not an overall verdict

Your workload, budget, and required context length matter most. A coding benchmark says little about visual understanding or agent performance.

The standard price table uses direct developer rates. The provider comparison labels OpenRouter endpoints, regions, and special tariffs separately.

Every result links to its measurement source and retrieval date. Labels distinguish vendor reports, official benchmark leaderboards, and independent evaluations. Disputed or archived results are excluded.

More matchups

Compare GPT-5.4 mini and Gemini 3.8 Flash with other leading models using the same data and benchmark logic.

View the complete LLM comparison

GPT-5.4 mini vs. Gemini 3.8 Flash

Change model selection

General specifications

API pricing compared

Shared benchmarks

Capability profile from matched benchmarks

Our tests

SVG short test

GPT-5.4 mini

Gemini 3.8 Flash

Pelican on a bicycle

Winning one benchmark is not an overall verdict

Related LLM comparisons